Deep Reinforcement Learning 2.0

Course Materials

Download the course materials here.

Description

In this course, you’ll explore and build an advanced AI model known as Twin-Delayed DDPG—a powerful algorithm that blends cutting-edge techniques in Artificial Intelligence, including continuous Double Deep Q-Learning, Policy Gradient, and Actor-Critic methods. This model is exceptionally robust—so much so that, for the first time in our series, we successfully train complex virtual agents like an ant, spider, and half-humanoid to walk and run across diverse terrains.

To help you master this model effectively, we’ve divided the course into three comprehensive parts:

Part 1: Fundamentals

In this section, we’ll cover the essential foundations of Artificial Intelligence that you need to fully grasp the AI concepts in this course. You’ll learn core techniques such as Q-Learning, Deep Q-Learning, Policy Gradient, Actor-Critic, and more. These fundamental topics will give you the knowledge and confidence to move forward with the advanced model in later parts.

Part 2: The Twin-Delayed DDPG Theory

Here, we’ll take a deep dive into the theory behind the Twin-Delayed DDPG model. Through a series of clear and insightful visualizations, you’ll follow the complete design and training process of this powerful AI. Not only will you gain a detailed understanding of the model’s architecture, but you’ll also develop strong intuition about how it learns and makes decisions. The combination of foundational knowledge from Part 1 and the in-depth theoretical insights in Part 2 will make this advanced model truly accessible—and by the end, you’ll be among the few who can master it.

Part 3: Twin-Delayed DDPG Implementation

In the final part, we’ll build the model from scratch together, step by step. Through interactive sessions and hands-on coding exercises—a new feature in this course—you’ll actively participate in the implementation process, reinforcing your skills as you go. Best of all, we’ll be working on Google Colab (Colaboratory), a free and open-source AI development platform. No installations are required—just hit “run,” and your AI will begin training. By the end, you’ll see your spider and humanoid agents running across the screen, powered by your very own code.